1 /* 2 * Copyright (c) 2008 ARM Ltd 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. The name of the company may not be used to endorse or promote 14 * products derived from this software without specific prior written 15 * permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS'' AND ANY EXPRESS OR IMPLIED 18 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF 19 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 20 * IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 21 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED 22 * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 23 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 24 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 25 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 26 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 27 */ 28 29 /* For GLIBC: 30 #include <string.h> 31 #include <memcopy.h> 32 33 #undef strcmp 34 */ 35 36 #ifdef __thumb2__ 37 #define magic1(REG) "#0x01010101" 38 #define magic2(REG) "#0x80808080" 39 #else 40 #define magic1(REG) #REG 41 #define magic2(REG) #REG ", lsl #7" 42 #endif 43 44 char* __attribute__((naked)) 45 strcpy (char* dst, const char* src) 46 { 47 asm ( 48 #if !(defined(__OPTIMIZE_SIZE__) || defined (PREFER_SIZE_OVER_SPEED) || \ 49 (defined (__thumb__) && !defined (__thumb2__))) 50 "pld [r1, #0]\n\t" 51 "eor r2, r0, r1\n\t" 52 "mov ip, r0\n\t" 53 "tst r2, #3\n\t" 54 "bne 4f\n\t" 55 "tst r1, #3\n\t" 56 "bne 3f\n" 57 "5:\n\t" 58 #ifndef __thumb2__ 59 "str r5, [sp, #-4]!\n\t" 60 "mov r5, #0x01\n\t" 61 "orr r5, r5, r5, lsl #8\n\t" 62 "orr r5, r5, r5, lsl #16\n\t" 63 #endif 64 65 "str r4, [sp, #-4]!\n\t" 66 "tst r1, #4\n\t" 67 "ldr r3, [r1], #4\n\t" 68 "beq 2f\n\t" 69 "sub r2, r3, "magic1(r5)"\n\t" 70 "bics r2, r2, r3\n\t" 71 "tst r2, "magic2(r5)"\n\t" 72 "itt eq\n\t" 73 "streq r3, [ip], #4\n\t" 74 "ldreq r3, [r1], #4\n" 75 "bne 1f\n\t" 76 /* Inner loop. We now know that r1 is 64-bit aligned, so we 77 can safely fetch up to two words. This allows us to avoid 78 load stalls. */ 79 ".p2align 2\n" 80 "2:\n\t" 81 "pld [r1, #8]\n\t" 82 "ldr r4, [r1], #4\n\t" 83 "sub r2, r3, "magic1(r5)"\n\t" 84 "bics r2, r2, r3\n\t" 85 "tst r2, "magic2(r5)"\n\t" 86 "sub r2, r4, "magic1(r5)"\n\t" 87 "bne 1f\n\t" 88 "str r3, [ip], #4\n\t" 89 "bics r2, r2, r4\n\t" 90 "tst r2, "magic2(r5)"\n\t" 91 "itt eq\n\t" 92 "ldreq r3, [r1], #4\n\t" 93 "streq r4, [ip], #4\n\t" 94 "beq 2b\n\t" 95 "mov r3, r4\n" 96 "1:\n\t" 97 #ifdef __ARMEB__ 98 "rors r3, r3, #24\n\t" 99 #endif 100 "strb r3, [ip], #1\n\t" 101 "tst r3, #0xff\n\t" 102 #ifdef __ARMEL__ 103 "ror r3, r3, #8\n\t" 104 #endif 105 "bne 1b\n\t" 106 "ldr r4, [sp], #4\n\t" 107 #ifndef __thumb2__ 108 "ldr r5, [sp], #4\n\t" 109 #endif 110 "BX LR\n" 111 112 /* Strings have the same offset from word alignment, but it's 113 not zero. */ 114 "3:\n\t" 115 "tst r1, #1\n\t" 116 "beq 1f\n\t" 117 "ldrb r2, [r1], #1\n\t" 118 "strb r2, [ip], #1\n\t" 119 "cmp r2, #0\n\t" 120 "it eq\n" 121 "BXEQ LR\n" 122 "1:\n\t" 123 "tst r1, #2\n\t" 124 "beq 5b\n\t" 125 "ldrh r2, [r1], #2\n\t" 126 #ifdef __ARMEB__ 127 "tst r2, #0xff00\n\t" 128 "iteet ne\n\t" 129 "strneh r2, [ip], #2\n\t" 130 "lsreq r2, r2, #8\n\t" 131 "streqb r2, [ip]\n\t" 132 "tstne r2, #0xff\n\t" 133 #else 134 "tst r2, #0xff\n\t" 135 "itet ne\n\t" 136 "strneh r2, [ip], #2\n\t" 137 "streqb r2, [ip]\n\t" 138 "tstne r2, #0xff00\n\t" 139 #endif 140 "bne 5b\n\t" 141 "BX LR\n" 142 143 /* src and dst do not have a common word-alignement. Fall back to 144 byte copying. */ 145 "4:\n\t" 146 "ldrb r2, [r1], #1\n\t" 147 "strb r2, [ip], #1\n\t" 148 "cmp r2, #0\n\t" 149 "bne 4b\n\t" 150 "BX LR" 151 152 #elif !defined (__thumb__) || defined (__thumb2__) 153 "mov r3, r0\n\t" 154 "1:\n\t" 155 "ldrb r2, [r1], #1\n\t" 156 "strb r2, [r3], #1\n\t" 157 "cmp r2, #0\n\t" 158 "bne 1b\n\t" 159 "BX LR" 160 #else 161 "mov r3, r0\n\t" 162 "1:\n\t" 163 "ldrb r2, [r1]\n\t" 164 "add r1, r1, #1\n\t" 165 "strb r2, [r3]\n\t" 166 "add r3, r3, #1\n\t" 167 "cmp r2, #0\n\t" 168 "bne 1b\n\t" 169 "BX LR" 170 #endif 171 ); 172 } 173 /* For GLIBC: libc_hidden_builtin_def (strcpy) */ 174